%%HTML
<script src="require.js"></script>

from IPython.display import HTML
HTML('''<script src="https://cdnjs.cloudflare.com/ajax/libs/jquery/2.0.3/jquery.min.js "></script><script>
code_show=true; 
function code_toggle() {
if (code_show){
$('div.jp-CodeCell > div.jp-Cell-inputWrapper').hide();
} else {
$('div.jp-CodeCell > div.jp-Cell-inputWrapper').show();
}
code_show = !code_show
} 
$( document ).ready(code_toggle);</script><form action="javascript:code_toggle()"><input type="submit" value="Toggle on/off for raw code"></form>
''')

# IMPORT LIBRARIES
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.svm import SVC
from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, classification_report, confusion_matrix
from imblearn.pipeline import Pipeline
from imblearn.over_sampling import RandomOverSampler
from sklearn.preprocessing import RobustScaler
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import GridSearchCV
from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier
from sklearn.preprocessing import LabelEncoder
import shap
from lime.lime_tabular import LimeTabularExplainer
from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import warnings


# Suppress warnings
warnings.filterwarnings('ignore')

# Map country names to country codes
country_codes_2023 = {
    'Argentina': 'ARG',
    'Brazil': 'BRA',
    'Bulgaria': 'BUL',
    'Canada': 'CAN',
    'China': 'CHN',
    'Cuba': 'CUB',
    'France': 'FRA',
    'Germany': 'GER',
    'Iran': 'IRI',
    'Italy': 'ITA',
    'Japan': 'JPN',
    'Netherlands': 'NED',
    'Poland': 'POL',
    'Serbia': 'SRB',
    'Slovenia': 'SLO',
    'USA': 'USA'
}

df_standings = pd.read_csv("data/VNL_2023_Standings.csv")

# replace with country codes for uniformity
df_standings['Team'] = df_standings['Team'].replace(country_codes_2023)
df_display = df_standings.head()
df_display = df_display.style.set_caption("<h4><b>Table 4.0: Volleyball Nations League 2023 Standings</b></h4>")
df_display.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '30px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '30px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])

df_standings_desc = pd.DataFrame(df_standings.describe(include='all').T)
df_standings_desc = df_standings_desc.style.set_caption("<h4><b>Table 5.0: Summary Statistics of Volleyball 2023 Standings Dataset</b></h4>")
df_standings_desc.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
df_standings_desc

types = pd.DataFrame(df_standings.dtypes)
types.columns = ['Type']
types = types.style.set_caption("<h4><b>Table 5.1: Volleyball 2023 Standings Dataset - Data Types</b></h4>")
types.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
types

null_counts = pd.DataFrame(df_standings.isnull().sum())
null_counts.columns = ['Count']
null_counts = null_counts.style.set_caption("<h4><b>Table 5.2: Volleyball 2023 Standings Dataset - Null Values Counts</b></h4>")
null_counts.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
null_counts

top_teams = df_standings[['Team', 'Points']].sort_values(by='Points', ascending=False).head(5)
top_teams = top_teams.style.set_caption("<h4><b>Table 5.3: Top 5 Teams based on Points</b></h4>")
top_teams.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '100px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '100px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
top_teams

# Create subplots with one row and three columns
fig, axes = plt.subplots(1, 3, figsize=(15, 5))

# Scatter plot for Matches Won vs Points
sns.scatterplot(x='Matches_Won', y='Points', data=df_standings, ax=axes[0])
axes[0].set_title('Matches Won vs Points')
axes[0].set_xlabel('Matches Won')
axes[0].set_ylabel('Points')

# Scatter plot for Sets Won vs Points
sns.scatterplot(x='Sets_Won', y='Points', data=df_standings, ax=axes[1])
axes[1].set_title('Sets Won vs Points')
axes[1].set_xlabel('Sets Won')
axes[1].set_ylabel('Points')

# Scatter plot for Points Won vs Point Ratio
sns.scatterplot(x='Points_Won', y='Point_Ratio', data=df_standings, ax=axes[2])
axes[2].set_title('Points Won vs Point Ratio')
axes[2].set_xlabel('Points Won')
axes[2].set_ylabel('Point Ratio')

# Adjust layout
plt.tight_layout()

# Add title to the entire diagram
plt.suptitle('Figure 1: Correlation Plots on Features', y=1.05, fontsize=14, fontweight='semibold')

# Show the plot
plt.show()

df_rosters = pd.read_csv("data/df_mens_rosters_21_23.csv")
df_rosters = df_rosters[df_rosters['Year'] == 2023]
df_rosters = df_rosters.drop(columns=['Year'])
df_display = df_rosters.head()
df_display = df_display.style.set_caption("<h4><b>Table 6.0: Volleyball 2023 Rosters (Men)</b></h4>")
df_display.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])

df_rosters_desc = pd.DataFrame(df_rosters.describe(include='all').T)
df_rosters_desc = df_rosters_desc.style.set_caption("<h4><b>Table 6.1: Summary Statistics of Volleyball Rosters 2023 Dataset</b></h4>")
df_rosters_desc.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
df_rosters_desc

types = pd.DataFrame(df_rosters.dtypes)
types.columns = ['Type']
types = types.style.set_caption("<h4><b>Table 6.2: Volleyball Rosters 2023 Dataset - Data Types</b></h4>")
types.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
types

# Converting to float
columns_to_float = [
    'No.', 'Age', 'Height', 'Total Points', 'Average by Match', 'Attack Points', 
    'Attack Avg Points', 'Block Points', 'Block Avg Points', 'Serve Points', 'Serve Avg Points'
]

for col in columns_to_float:
    df_rosters[col] = pd.to_numeric(df_rosters[col], errors='coerce')

# Converting Player_ID to object (string)
df_rosters['Player_ID'] = df_rosters['Player_ID'].astype(str)

# Removing '%' and converting to float for specific columns
percentage_columns = ['Efficiency', 'Block Success', 'Serve Success']
for col in percentage_columns:
    df_rosters[col] = df_rosters[col].replace('-', '0')  # Replace '-' with '0'
    df_rosters[col] = df_rosters[col].str.rstrip('%').astype(float) / 100

types = pd.DataFrame(df_rosters.dtypes)
types.columns = ['Type']
types = types.style.set_caption("<h4><b>Table 6.3: Volleyball Rosters 2023 Dataset - Fixed Data Types</b></h4>")
types.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
types

df_rosters_desc = pd.DataFrame(df_rosters.describe().T)
df_rosters_desc = df_rosters_desc.style.set_caption("<h4><b>Table 6.4: Fixed Summary Statistics of Volleyball Rosters 2023 Dataset</b></h4>")
df_rosters_desc.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
df_rosters_desc

null_counts = pd.DataFrame(df_rosters.isnull().sum())
null_counts.columns = ['Count']
null_counts = null_counts.style.set_caption("<h4><b>Table 6.5: Volleyball Rosters 2023 Dataset - Null Values Counts</b></h4>")
null_counts.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '200px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
null_counts

df_rosters_null_desc = df_rosters[df_rosters['No.'].isnull()]
#df_rosters[df_rosters['No.'].isnull()]
df_rosters_null_desc = df_rosters_null_desc.style.set_caption("<h4><b>Table 6.6: Volleyball Rosters 2023 Dataset - Null Data</b></h4>")
df_rosters_null_desc.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])
df_rosters_null_desc

df_rosters = df_rosters[df_rosters['Position'] != 'COACH']

df_rosters.fillna(0, inplace=True)

df_filtered = df_rosters.drop(columns=[
    'No.', 'Player Name', 'Player_ID', 'Nationality'
])

# Iterate over each country and plot histograms for numerical columns
countries = df_filtered['Country_Name'].unique()
for index, country in enumerate(countries):
    # Filter dataframe for the current country
    df_country = df_filtered[df_filtered['Country_Name'] == country]
    
    # Select numerical columns
    numerical_columns = df_country.select_dtypes(include=['float64', 'int64']).columns
    
    # Initialize figure
    plt.figure(figsize=(12, 8))
    
    # Plot histograms for each numerical column
    for i, col in enumerate(numerical_columns):
        plt.subplot(4, 4, i + 1)
        sns.histplot(df_country[col], kde=True, color='skyblue')
        plt.title(col)
        plt.xlabel('')
        plt.ylabel('Frequency')
    
    # Adjust index by adding 1 (for human-readable figure numbering)
    figure_number = index + 2.0
    
    # Set the suptitle with dynamic figure number and country name
    plt.suptitle(f'Figure {figure_number}: Player Profiles for {country}', fontsize=12, fontweight='semibold')
    
    # Save and show the figure
    plt.tight_layout()
    plt.show()

# Drop irrelevant columns
df_filtered = df_rosters.drop(columns=[
    'No.', 'Player Name', 'Player_ID', 'Nationality'
])

# Group by 'Country_Name'
grouped = df_filtered.groupby('Country_Name')

# Aggregate numerical data by mean and standard deviation
numerical_stats = grouped.mean().add_prefix('Mean_Player_').join(grouped.std().add_prefix('Std_Player_'))
df_team_profile = numerical_stats.copy()


# Reset index
df_team_profile.reset_index(inplace=True)

# Replace country names with country codes
df_team_profile['Country_Name']= df_team_profile['Country_Name'].replace(country_codes_2023)

df_display = df_team_profile
df_display = df_display.style.set_caption("<h4><b>Table 7.0: Volleyball 2023 Team Country Profiles</b></h4>")
df_display.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])

df_matches = pd.read_csv('data/vnl_matches_2023.csv')

# Only include matches from the current roster
df_matches = df_matches[(df_matches['Team1'].isin(country_codes_2023.values())) & (df_matches['Team2'].isin(country_codes_2023.values()))]
df_matches.rename(columns={'Team1_Points': 'Team1_Performance_Ranking', 'Team2_Points': 'Team2_Performance_Ranking'}, inplace=True)

df_display = df_matches.head()
df_display = df_display.style.set_caption("<h4><b>Table 8.0: Volleyball 2023 Matches</b></h4>")
df_display.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '100px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '100px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])

# df_matches['Results'].value_counts()

# Add suffixes to the columns of df_team_profile
df_team_profile_team1 = df_team_profile.add_prefix('Team1_')
df_team_profile_team2 = df_team_profile.add_prefix('Team2_')

# Merge df_matches with df_team_profile for Team 1
df_agg_matches = pd.merge(df_matches, df_team_profile_team1, left_on='Team1', right_on='Team1_Country_Name', how='left')

# Merge df_matches with df_team_profile for Team 2
df_agg_matches = pd.merge(df_agg_matches, df_team_profile_team2, left_on='Team2', right_on='Team2_Country_Name', how='left')

df_agg_matches.drop(columns=['Team1_Country_Name', 'Team2_Country_Name'], inplace=True)

# Get column names starting with 'Team1_'
team1_columns = [col for col in df_agg_matches.columns if col.startswith('Team1')]

# Get column names starting with 'Team2_'
team2_columns = [col for col in df_agg_matches.columns if col.startswith('Team2')]

# Reorder columns
new_columns_order = team1_columns + team2_columns + ['Results']
df_agg_matches = df_agg_matches[new_columns_order]

df_display = df_agg_matches[0:14]
df_display = df_display.style.set_caption("<h4><b>Table 9.0: Aggregated Matches</b></h4>")
df_display.set_table_styles([{
    'selector': 'th',
    'props': [('text-align', 'center'), ('min-width', '100px')]
}, {
    'selector': 'td',
    'props': [('text-align', 'center'), ('min-width', '100px')]
}, {
    'selector': '',
    'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
}])

# Define the parameter grid
param_grid = [
    {
        'clf': [DecisionTreeClassifier()],
        'clf__max_depth': [2, 3, 4]
    },
    {
        'clf': [RandomForestClassifier()],
        'clf__max_depth': [2, 3, 4, 5],
        'clf__n_estimators': [10, 20, 50, 100, 150, 200]
    },
    {
        'clf': [GradientBoostingClassifier()],
        'clf__max_depth': [2, 3, 4, 5],
        'clf__n_estimators': [10, 20, 50, 100, 150, 200]
    },
    {
        'clf': [KNeighborsClassifier()],
        'clf__n_neighbors': [2, 3, 4, 5, 6, 7, 8, 9]
    },
    {
        'clf': [SVC()],
        'clf__kernel': ['rbf', 'sigmoid', 'poly'],
        'clf__C': [0.001, 0.05, 0.01, 0.5, 0.1, 1, 10]
    },
]

df_win_loss = df_agg_matches.copy()

def convert_to_win_loss(result):
    team1_sets, team2_sets = map(int, result.split(' - '))
    return '1 - 0' if team1_sets > team2_sets else '0 - 1'

df_win_loss['Results_2'] = df_win_loss['Results'].apply(convert_to_win_loss)
df_win_loss.drop(columns=['Results'], inplace=True)

label_encoder_win_loss = LabelEncoder()
target = 'Results_2'

df_win_loss_encoded = df_win_loss.copy()
X_win_loss = df_win_loss_encoded.copy()
X_win_loss.drop(columns=['Team1', 'Team2', target], inplace=True)
y_win_loss = label_encoder_win_loss.fit_transform(df_win_loss_encoded[target])

# Get the label encoder dictionary
result_label_dict_win_loss = dict(zip(label_encoder_win_loss.transform(label_encoder_win_loss.classes_), label_encoder_win_loss.classes_))

X_train_win_loss, X_test_win_loss, y_train_win_loss, y_test_win_loss = train_test_split(X_win_loss, y_win_loss, test_size=0.2, random_state=42)

pipe_win_loss = Pipeline([
    ('scaler', RobustScaler()),
    ('clf', None)
])

grid_search_win_loss = GridSearchCV(
    estimator=pipe_win_loss,
    param_grid=param_grid,
    cv=10,
    scoring='accuracy',
    n_jobs=-1,
    return_train_score=True
)

# COMMENTED THIS OUT - Note that there will be some randomness on the estimators and max_depth
# grid_search_win_loss.fit(X_train_win_loss, y_train_win_loss)
# print(f"The best model is: {grid_search_win_loss.best_params_}")
# print(f"with an accuracy score of: {grid_search_win_loss.best_score_}")

# results_win_loss = grid_search_win_loss.cv_results_
# scores_df_win_loss = pd.DataFrame(results_win_loss)
# scores_df_win_loss.sort_values(by='rank_test_score').head(10)

pipe_win_loss = Pipeline([
    ('scaler', RobustScaler()),
    ('classifier', GradientBoostingClassifier(max_depth=4, n_estimators=100, random_state=1337))
])

# Train the model using the training-validation dataset
pipe_win_loss.fit(X_train_win_loss, y_train_win_loss)

Pipeline(steps=[('scaler', RobustScaler()),
                ('classifier',
                 GradientBoostingClassifier(max_depth=4, random_state=1337))])

Pipeline(steps=[('scaler', RobustScaler()),
                ('classifier',
                 GradientBoostingClassifier(max_depth=4, random_state=1337))])

RobustScaler()

GradientBoostingClassifier(max_depth=4, random_state=1337)

y_pred_test_win_loss = pipe_win_loss.predict(X_test_win_loss)

precision_win_loss = precision_score(y_test_win_loss, y_pred_test_win_loss, average=None)
recall_win_loss = recall_score(y_test_win_loss, y_pred_test_win_loss, average=None)
f1_win_loss = f1_score(y_test_win_loss, y_pred_test_win_loss, average=None)
accuracy_win_loss = accuracy_score(y_test_win_loss, y_pred_test_win_loss)

print(f"Average Precision Score on the test set: {np.mean(precision_win_loss):.4f}")
print(f"Average Recall Score on the test set: {np.mean(recall_win_loss):.4f}")
print(f"Average F1 Score on the test set: {np.mean(f1_win_loss):.4f}")
print(f"Accuracy Score on the test set: {accuracy_win_loss:.4f}")

Average Precision Score on the test set: 0.9310
Average Recall Score on the test set: 0.9500
Average F1 Score on the test set: 0.9366
Accuracy Score on the test set: 0.9385

cm_win_loss = confusion_matrix(y_test_win_loss, y_pred_test_win_loss, labels=pipe_win_loss.classes_)
disp_win_loss = ConfusionMatrixDisplay(confusion_matrix=cm_win_loss,
                              display_labels=pipe_win_loss.classes_)
fig, ax = plt.subplots(figsize=(5, 5))
disp_win_loss.plot(ax=ax, colorbar=True);

fig.suptitle("Figure 18.0:Confusion Matrix for Win/Loss Prediction", fontsize=12, verticalalignment='top')
plt.show()

best_model_win_loss = pipe_win_loss.fit(X_win_loss, y_win_loss)

label_encoder = LabelEncoder()

target = 'Results'

df_agg_matches_encoded = df_agg_matches.copy()
X = df_agg_matches.copy()
X.drop(columns=['Team1', 'Team2', target], inplace=True)
y =  label_encoder.fit_transform(df_agg_matches_encoded[target])

result_label_dict = dict(zip(label_encoder.transform(label_encoder.classes_), label_encoder.classes_))
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=1337)

pipe_set = Pipeline([
    ('scaler', RobustScaler()),
    ('clf', None)
])

grid_search = GridSearchCV(
    estimator=pipe_set,
    param_grid=param_grid,
    cv=10,
    scoring='accuracy',
    n_jobs=-1,
    return_train_score=True
)

# COMMENTED THIS OUT - Note that there will be some randomness on the estimators and max_depth
# grid_search.fit(X_train, y_train)
# print(f"The best model is: {grid_search.best_params_}")
# print(f"with a accuracy score of: {grid_search.best_score_}")

# results = grid_search.cv_results_
# scores_df = pd.DataFrame(results)
# df_display = scores_df.sort_values(by='rank_test_score').head(10)
# df_display = df_display.style.set_caption("<h4><b>Table 11.0: Gridsearch Results for Best Model</b></h4>")
# df_display.set_table_styles([{
#     'selector': 'th',
#     'props': [('text-align', 'center'), ('min-width', '100px')]
# }, {
#     'selector': 'td',
#     'props': [('text-align', 'center'), ('min-width', '100px')]
# }, {
#     'selector': '',
#     'props': [('margin-left', 'auto'), ('margin-right', 'auto')]
# }])

pipe_set = Pipeline([
    ('scaler', RobustScaler()),
    ('classifier', GradientBoostingClassifier(max_depth=2, n_estimators=200, random_state=1337))
])

# Train the model using the training-validation dataset
pipe_set.fit(X_train, y_train)

Pipeline(steps=[('scaler', RobustScaler()),
                ('classifier',
                 GradientBoostingClassifier(max_depth=2, n_estimators=200,
                                            random_state=1337))])

Pipeline(steps=[('scaler', RobustScaler()),
                ('classifier',
                 GradientBoostingClassifier(max_depth=2, n_estimators=200,
                                            random_state=1337))])

RobustScaler()

GradientBoostingClassifier(max_depth=2, n_estimators=200, random_state=1337)

y_pred_test = pipe_set.predict(X_test)

precision = precision_score(y_test, y_pred_test, average=None)
recall = recall_score(y_test, y_pred_test, average=None)
f1 = f1_score(y_test, y_pred_test, average=None)
accuracy = accuracy_score(y_test, y_pred_test)

print(f"Average Precision Score on the test set: {np.mean(precision):.4f}")
print(f"Average Recall Score on the test set: {np.mean(recall):.4f}")
print(f"Average F1 Score on the test set: {np.mean(f1):.4f}")
print(f"Accuracy Score on the test set: {accuracy:.4f}")

Average Precision Score on the test set: 0.7380
Average Recall Score on the test set: 0.7125
Average F1 Score on the test set: 0.7167
Accuracy Score on the test set: 0.7320

cm_set = confusion_matrix(y_test, y_pred_test, labels=pipe_set.classes_)
disp_set = ConfusionMatrixDisplay(confusion_matrix=cm_set,
                              display_labels=pipe_set.classes_)

fig, ax = plt.subplots(figsize=(5, 5))  # Adjust the figsize as needed
disp_set.plot(ax=ax, colorbar=True);

fig.suptitle("Figure 19.0: Confusion Matrix for Set Score Prediction", fontsize=12, verticalalignment='top')
plt.show()

best_model_set = pipe_set.fit(X, y)

shap_explainer = shap.Explainer(pipe_set.predict_proba, X_train, feature_names=X_train.columns)
shap_values = shap_explainer(X_test)

PermutationExplainer explainer: 98it [00:51,  1.70it/s]

shap_explanation = shap.Explanation(shap_values.values[:, :, 1],
                                    shap_values.base_values[0][1],
                                    shap_values.data,
                                    feature_names=X_test.columns)

shap.summary_plot(shap_explanation, plot_type='bar', plot_size=(14, 8), show=False)

plt.title("Figure 20.0: SHAP Summary Plot", fontsize=16, fontweight='bold')
plt.show()

# Graphing Shap distribution for feature
top_15_features = np.abs(shap_explanation.values).mean(axis=0).argsort()[::-1][:15]

fig, axes = plt.subplots(5, 3, figsize=(15, 25))

# Iterate over each feature for rows
for i, feature in enumerate(top_15_features):
    row_index = i // 3
    col_index = i % 3
    shap.plots.scatter(shap_explanation[:, feature], ax=axes[row_index, col_index], show=False)
    axes[row_index, col_index].set_title(f"SHAP Values for {X.columns[feature]}")


plt.tight_layout(rect=[0, 0, 1, 0.95])
plt.suptitle('Figure 21: Top 15 Features', fontsize=16, fontweight='bold')
plt.show()

shap.summary_plot(shap_explanation, plot_size=(14, 8), show=False)

plt.title("Figure 22.0: SHAP Summary Plot", fontsize=16, fontweight='bold')
plt.show()

# Updated country names to country codes
country_codes_2024 = {
    'Argentina': 'ARG',
    'Brazil': 'BRA',
    'Bulgaria': 'BUL',
    'Canada': 'CAN',
    'Cuba': 'CUB',
    'France': 'FRA',
    'Germany': 'GER',
    'Iran': 'IRI',
    'Italy': 'ITA',
    'Japan': 'JPN',
    'Netherlands': 'NED',
    'Poland': 'POL',
    'Serbia': 'SRB',
    'Slovenia': 'SLO',
    'USA': 'USA',
    'Turkey': 'TUR'
}

df_2024_rosters = pd.read_csv('data/df_mens_rosters_24.csv')

# Converting to float
columns_to_float = [
    'No.', 'Age', 'Height', 'Total Points', 'Average by Match', 'Attack Points', 
    'Attack Avg Points', 'Block Points', 'Block Avg Points', 'Serve Points', 'Serve Avg Points'
]

for col in columns_to_float:
    df_2024_rosters[col] = pd.to_numeric(df_2024_rosters[col], errors='coerce')

# Converting Player_ID to object (string)
df_2024_rosters['Player_ID'] = df_2024_rosters['Player_ID'].astype(str)

# Removing '%' and converting to float for specific columns
percentage_columns = ['Efficiency', 'Block Success', 'Serve Success']
for col in percentage_columns:
    df_2024_rosters[col] = df_2024_rosters[col].replace('-', '0')  # Replace '-' with '0'
    df_2024_rosters[col] = df_2024_rosters[col].str.rstrip('%').astype(float) / 100

df_2024_rosters = df_2024_rosters[df_2024_rosters['Position'] != 'COACH']
df_2024_rosters.fillna(0, inplace=True)

# Drop irrelevant columns
df_filtered = df_2024_rosters.drop(columns=[
    'No.', 'Year', 'Player Name', 'Player_ID', 'Nationality'
])

# Group by 'Country_Name'
grouped = df_filtered.groupby('Country_Name')

# Aggregate numerical data by mean and standard deviation
numerical_stats = grouped.mean().add_prefix('Mean_Player_').join(grouped.std().add_prefix('Std_Player_'))
df_2024_team_profile = numerical_stats.copy() 

# Reset index
df_2024_team_profile.reset_index(inplace=True)

# Replace country names with country codes
df_2024_team_profile['Country_Name'] = df_2024_team_profile['Country_Name'].replace(country_codes_2024)

df_display = df_2024_team_profile

df_display.style.set_caption(
    "Table 10.0: 2024 Country Team Profiles"
    ).set_table_styles([{
    'selector': 'caption',
    'props': [('text-align', 'center'), 
              ('font-size', '110%'), 
              ('font-weight', 'bold')]
}])

# Get their current points
df_2023_standings = pd.read_csv('data/vnl_2023_current_standing.csv')

def createMatchRecord(team1, team2, team_profile):
    # Add suffixes to the columns of df_team_profile
    df_team_profile_team1 = team_profile[team_profile['Country_Name'] == team1].add_prefix('Team1_')
    df_team_profile_team2 = team_profile[team_profile['Country_Name'] == team2].add_prefix('Team2_')
    
    df_2023_standings_team1 = df_2023_standings[df_2023_standings['Team'] == team1].rename(columns={'Team': 'Team1', 'Points': 'Team1_Performance_Ranking'})
    df_2023_standings_team2 = df_2023_standings[df_2023_standings['Team'] == team2].rename(columns={'Team': 'Team2', 'Points': 'Team2_Performance_Ranking'})
    df_2023_standings_team1.index = [0]
    df_2023_standings_team2.index = [0]
    df_match = df_2023_standings_team1.merge(df_2023_standings_team2, how='outer', left_index=True, right_index=True)
    
    # Merge df_matches with df_team_profile for Team 1
    df_match = pd.merge(df_match, df_team_profile_team1, left_on='Team1', right_on='Team1_Country_Name', how='left')
    
    # Merge df_matches with df_team_profile for Team 2
    df_match = pd.merge(df_match, df_team_profile_team2, left_on='Team2', right_on='Team2_Country_Name', how='left')


    df_match.drop(columns=['Team1_Country_Name', 'Team2_Country_Name'], inplace=True)
    
    # Get column names starting with 'Team1_'
    team1_columns = [col for col in df_match.columns if col.startswith('Team1')]
    
    # Get column names starting with 'Team2_'
    team2_columns = [col for col in df_match.columns if col.startswith('Team2')]
    
    # Reorder columns
    new_columns_order = team1_columns + team2_columns 
    df_match = df_match[new_columns_order]
    df_match.drop(columns=['Team1', 'Team2'], inplace=True)
    return df_match, df_team_profile_team1, df_team_profile_team2


def predictMatchSet(match):
    result =  best_model_set.predict(match)
    return label_encoder.inverse_transform([result[0]])[0]


def predictMatchWin(match):
    result =  best_model_win_loss.predict(match)
    return label_encoder_win_loss.inverse_transform([result[0]])[0]


def getLabelClassSet(label):
    return int(label_encoder.transform([label])[0])


def getLabelClassWin(label):
    return int(label_encoder_win_loss.transform([label])[0])

team1 = 'CUB'
team2 = 'JPN'

match, team1_profile, team2_profile = createMatchRecord(team1, team2, df_2024_team_profile)

# Using set outcome model
set_outcome = predictMatchSet(match)
print(f'Predicted Set Outcome for Team 1 ({team1}) vs Team 2 ({team2}): {set_outcome}')
print(f'Predicted Set Winner: {team1 if getLabelClassSet(set_outcome) > 2 else team2}')

Predicted Set Outcome for Team 1 (CUB) vs Team 2 (JPN): 1 - 3
Predicted Set Winner: JPN

# Using Win-Loss Model
win_outcome = predictMatchWin(match)
print(f'Predicted Win Outcome for Team 1 ({team1}) vs Team 2 ({team2}): {team1 if getLabelClassWin(win_outcome) == 1 else team2}')

Predicted Win Outcome for Team 1 (CUB) vs Team 2 (JPN): JPN

shap_explainer_match = shap.Explainer(
    best_model_set.predict_proba, X, feature_names=X.columns)

shap_values_match = shap_explainer_match(match)

shap_explanation_match = shap.Explanation(
    shap_values_match.values[:, :, 1],
    shap_values_match.base_values[0][1],
    shap_values_match.data,
    feature_names=X.columns
)

shap.plots.waterfall(shap_explanation_match[0], show=False)

plt.title("Figure 23.0: SHAP Waterfall Plot", fontsize=12, fontweight='bold')
plt.show()

lime_explainer = LimeTabularExplainer(
    X.to_numpy(),
    feature_names=X.columns
)
lime_explanation = lime_explainer.explain_instance(
    match.iloc[0],
    best_model_set.predict_proba,
    num_features=10
)

lime_html = lime_explanation.as_html()

custom_html = f"""
<div style='text-align: center; margin-bottom: 20px;'>
    <<h2 style='font-size: 16px; 'font-weight: bold'>Figure 22.0: LIME Explanation for Features</h2>
</div>
{lime_html}
"""

# Display the custom HTML in the notebook
display(HTML(custom_html))

def test_predict_with_explanation(team1, team2, figure_number):
    match, team1_profile, team2_profile = createMatchRecord(team1, team2, df_2024_team_profile)
    win_outcome = predictMatchWin(match)
    set_outcome = predictMatchSet(match)
    
    print(f'Predicted Win Model Outcome for Team 1 ({team1}) vs Team 2 ({team2}): {team1 if getLabelClassWin(win_outcome) == 1 else team2}')
    print(f'Predicted Set Model Outcome for Team 1 ({team1}) vs Team 2 ({team2}): {set_outcome} ({team1 if getLabelClassSet(set_outcome) > 2 else team2})')
    
    lime_explanation_match = lime_explainer.explain_instance(
        match.iloc[0],
        best_model_set.predict_proba,
        num_features=5
    )
    
    lime_html = lime_explanation_match.as_html(show_all=False)

    # Create custom HTML with a dynamic title based on parameters
    custom_html = f"""
    <div style='text-align: center; margin-bottom: 20px;'>
        <h2 style='font-size: 18px; font-weight: bold;'>Figure {figure_number}: Explanation for Features - {team1} vs {team2}</h2>
    </div>
        <div style='text-align: center; margin-bottom: 20px;'>
        <h3 style='font-size: 16px; font-weight: semibold;'>LIME Explanation for Top 5 Features</h2>
    </div>
    {lime_html}

     </div>
        <div style='text-align: center; margin-top: 20px;'>
        <h3 style='font-size: 16px; font-weight: semibold;'>SHAP Explanation Top 5 Features</h2>
    </div>
    """

    # Display the custom HTML in the notebook
    display(HTML(custom_html))

     # SHAP explanation
    shap_explainer_match = shap.Explainer(best_model_set.predict_proba, X, feature_names=X.columns)
    
    shap_values_match = shap_explainer_match(match)

    shap_explanation_match = shap.Explanation(
        shap_values_match.values[:, :, 1],
        shap_values_match.base_values[0][1],
        shap_values_match.data,
        feature_names=X.columns
    )
    shap.plots.waterfall(shap_explanation_match[0],  max_display=6, show=False)
    plt.tight_layout()
    plt.show()

team1 = 'ARG'
team2 = 'USA'
figure_number = '25.0'

test_predict_with_explanation(team1, team2, figure_number)

Predicted Win Model Outcome for Team 1 (ARG) vs Team 2 (USA): ARG
Predicted Set Model Outcome for Team 1 (ARG) vs Team 2 (USA): 3 - 0 (ARG)

team1 = 'CAN'
team2 = 'CUB'
figure_number = '26.0'

test_predict_with_explanation(team1, team2, figure_number)

Predicted Win Model Outcome for Team 1 (CAN) vs Team 2 (CUB): CUB
Predicted Set Model Outcome for Team 1 (CAN) vs Team 2 (CUB): 3 - 0 (CAN)

team1 = 'SLO'
team2 = 'TUR'
figure_number = '27.0'

test_predict_with_explanation(team1, team2, figure_number)

Predicted Win Model Outcome for Team 1 (SLO) vs Team 2 (TUR): SLO
Predicted Set Model Outcome for Team 1 (SLO) vs Team 2 (TUR): 3 - 0 (SLO)

team1 = 'GER'
team2 = 'JPN'
figure_number = '28.0'

test_predict_with_explanation(team1, team2, figure_number)

Predicted Win Model Outcome for Team 1 (GER) vs Team 2 (JPN): GER
Predicted Set Model Outcome for Team 1 (GER) vs Team 2 (JPN): 3 - 2 (GER)

Metric	Bet Smart (Win-Lose Model)	Bet Smarter (Set-Score Model)
Accuracy	93.85%	73.20%
Precision	93.10%	73.80%
Recall	95.00%	71.25%
F1 Score	93.66%	71.67%

Feature Name	Description	Data Type
Rank	The position of the team in the league standings.	Integer
Team	The name of the volleyball team.	Text
Matches_Total	The total number of matches played by the team.	Integer
Matches_Won	The number of matches won by the team.	Integer
Matches_Lost	The number of matches lost by the team.	Integer
3-0	The number of matches won by the team with a score of 3-0.	Integer
3-1	The number of matches won by the team with a score of 3-1.	Integer
3-2	The number of matches won by the team with a score of 3-2.	Integer
2-3	The number of matches lost by the team with a score of 2-3.	Integer
1-3	The number of matches lost by the team with a score of 1-3.	Integer
0-3	The number of matches lost by the team with a score of 0-3.	Integer
Points	The total points accumulated by the team.	Integer
Sets_Won	The total number of sets won by the team.	Integer
Sets_Lost	The total number of sets lost by the team.	Integer
Set_Ratio	The ratio of sets won to sets lost.	Float
Points_Won	The total number of points won by the team.	Integer
Points_Lost	The total number of points lost by the team.	Integer
Point_Ratio	The ratio of points won to points lost.	Float

Feature Name	Description	Data Type
No.	Jersey number of the player	Integer
Player Name	Full name of the player	Text
Position	Position played by the player on the team	Text
Player_ID	Unique identifier for each player	Integer
Year	Year of the roster entry	Integer
Country_Name	Country that the player represented	Text
Nationality	Nationality of the player	Text
Age	Age of the player	Integer
Height	Height of the player in centimeters	Integer
Total Points	Total points scored by the player	Text
Average by Match	Average points scored by the player per match	Text
Attack Points	Points scored by the player through attacks	Text
Efficiency	Attack efficiency percentage	Text
Attack Avg Points	Average attack points scored per match	Text
Block Points	Points scored by the player through blocks	Text
Block Success	Success percentage of blocks attempted	Text
Block Avg Points	Average block points scored per match	Text
Serve Points	Points scored by the player through serves	Text
Serve Success	Success percentage of serves attempted	Text
Serve Avg Points	Average serve points scored per match	Text

Feature Name	Description	Data Type
Match_ID	Unique identifier for each match	Integer
Date	Date when the match was played	Date
Team_A	Name of one competing team	Text
Team_B	Name of the opposing team	Text
Set_Score	Score by sets, e.g., 3-2, showing sets won by each team	Text
Winning_Team	Name of the team that won the match	Text
Venue	Location where the match was played	Text

	count	unique	top	freq	mean	std	min	25%	50%	75%	max
Rank	16.000000	nan	nan	nan	8.500000	4.760952	1.000000	4.750000	8.500000	12.250000	16.000000
Team	16	16	USA	1	nan	nan	nan	nan	nan	nan	nan
Matches_Total	16.000000	nan	nan	nan	12.000000	0.000000	12.000000	12.000000	12.000000	12.000000	12.000000
Matches_Won	16.000000	nan	nan	nan	6.000000	3.162278	2.000000	3.000000	6.000000	9.000000	10.000000
Matches_Lost	16.000000	nan	nan	nan	6.000000	3.162278	2.000000	3.000000	6.000000	9.000000	10.000000
3-0	16.000000	nan	nan	nan	2.500000	2.160247	0.000000	1.000000	3.000000	3.000000	9.000000
3-1	16.000000	nan	nan	nan	2.125000	1.586401	0.000000	1.000000	1.500000	3.250000	5.000000
3-2	16.000000	nan	nan	nan	1.375000	1.454877	0.000000	0.000000	1.000000	2.250000	5.000000
2-3	16.000000	nan	nan	nan	1.375000	1.408309	0.000000	0.000000	1.000000	2.000000	5.000000
1-3	16.000000	nan	nan	nan	2.125000	1.821172	0.000000	1.000000	1.000000	4.000000	5.000000
0-3	16.000000	nan	nan	nan	2.500000	2.000000	0.000000	1.000000	2.000000	4.250000	6.000000
Points	16.000000	nan	nan	nan	18.000000	8.438009	6.000000	9.750000	17.500000	25.250000	31.000000
Sets_Won	16.000000	nan	nan	nan	22.875000	7.446476	12.000000	15.750000	23.000000	30.000000	33.000000
Sets_Lost	16.000000	nan	nan	nan	22.875000	7.787811	7.000000	17.750000	22.000000	31.000000	33.000000
Set_Ratio	16.000000	nan	nan	nan	1.308125	1.076667	0.363000	0.507750	1.047500	1.693750	4.714000
Points_Won	16.000000	nan	nan	nan	1033.000000	66.260094	941.000000	985.250000	1025.500000	1061.500000	1184.000000
Points_Lost	16.000000	nan	nan	nan	1033.000000	64.230834	871.000000	1026.500000	1049.500000	1073.250000	1119.000000
Point_Ratio	16.000000	nan	nan	nan	1.002875	0.082851	0.876000	0.924000	1.013500	1.072000	1.132000

	Rank	Team	Matches_Total	Matches_Won	Matches_Lost	3-0	3-1	3-2	2-3	1-3	0-3	Points	Sets_Won	Sets_Lost	Set_Ratio	Points_Won	Points_Lost	Point_Ratio
0	1	USA	12	10	2	9	1	0	1	1	0	31	33	7	4.714000	986	871	1.132000
1	2	JPN	12	10	2	3	4	3	0	1	1	27	31	16	1.937000	1109	1034	1.072000
2	3	POL	12	10	2	2	3	5	0	0	2	25	30	19	1.578000	1108	1057	1.048000
3	4	ITA	12	9	3	4	4	1	0	1	2	26	28	15	1.866000	1016	906	1.121000
4	5	ARG	12	9	3	3	3	3	2	1	0	26	32	18	1.777000	1184	1092	1.084000

	Type
Rank	int64
Team	object
Matches_Total	int64
Matches_Won	int64
Matches_Lost	int64
3-0	int64
3-1	int64
3-2	int64
2-3	int64
1-3	int64
0-3	int64
Points	int64
Sets_Won	int64
Sets_Lost	int64
Set_Ratio	float64
Points_Won	int64
Points_Lost	int64
Point_Ratio	float64

	No.	Player Name	Position	Player_ID	Country_Name	Nationality	Age	Height	Total Points	Average by Match	Attack Points	Efficiency	Attack Avg Points	Block Points	Block Success	Block Avg Points	Serve Points	Serve Success	Serve Avg Points
701	1.000000	Sanchez Matias	S	142412.000000	Argentina	Argentina	27.000000	175.000000	1	0.08	0	0.00%	0.00	0	0.00%	0.00	1	1.61%	0.08
702	3.000000	Martinez Franchi Jan	U	142410.000000	Argentina	Argentina	26.000000	190.000000	28	2.15	25	38.46%	1.92	2	12.50%	0.15	1	2.86%	0.08
703	4.000000	Gallego Joaquin	MB	142411.000000	Argentina	Argentina	27.000000	204.000000	20	1.54	14	46.67%	1.08	6	19.35%	0.46	0	0.00%	0.00
704	7.000000	Palonsky Luciano	OH	162546.000000	Argentina	Argentina	24.000000	198.000000	183	14.08	151	49.35%	11.62	13	18.84%	1.00	19	9.31%	1.46
705	8.000000	Loser Agustin	MB	152692.000000	Argentina	Argentina	26.000000	198.000000	168	12.92	103	63.98%	7.92	53	32.52%	4.08	12	5.56%	0.92

	count	unique	top	freq	mean	std	min	25%	50%	75%	max
No.	224.000000	nan	nan	nan	13.941964	13.199444	1.000000	6.000000	12.000000	18.000000	99.000000
Player Name	240	240	Sanchez Matias	1	nan	nan	nan	nan	nan	nan	nan
Position	240	7	OH	74	nan	nan	nan	nan	nan	nan	nan
Player_ID	224.000000	nan	nan	nan	151379.616071	19832.302777	113691.000000	137261.750000	152387.500000	168268.250000	192453.000000
Country_Name	240	16	Argentina	15	nan	nan	nan	nan	nan	nan	nan
Nationality	224	16	Argentina	14	nan	nan	nan	nan	nan	nan	nan
Age	224.000000	nan	nan	nan	27.691964	4.354136	19.000000	25.000000	27.000000	30.000000	41.000000
Height	224.000000	nan	nan	nan	196.200893	15.602685	2.000000	192.000000	198.000000	203.250000	215.000000
Total Points	224	108	0	36	nan	nan	nan	nan	nan	nan	nan
Average by Match	224	151	0.00	30	nan	nan	nan	nan	nan	nan	nan
Attack Points	224	98	0	48	nan	nan	nan	nan	nan	nan	nan
Efficiency	224	148	-	40	nan	nan	nan	nan	nan	nan	nan
Attack Avg Points	224	132	-	40	nan	nan	nan	nan	nan	nan	nan
Block Points	224	28	0	56	nan	nan	nan	nan	nan	nan	nan
Block Success	224	120	-	36	nan	nan	nan	nan	nan	nan	nan
Block Avg Points	224	74	-	36	nan	nan	nan	nan	nan	nan	nan
Serve Points	224	23	0	77	nan	nan	nan	nan	nan	nan	nan
Serve Success	224	121	0.00%	40	nan	nan	nan	nan	nan	nan	nan
Serve Avg Points	224	60	0.00	40	nan	nan	nan	nan	nan	nan	nan

	No.	Player Name	Position	Player_ID	Country_Name	Nationality	Age	Height	Total Points	Average by Match	Attack Points	Efficiency	Attack Avg Points	Block Points	Block Success	Block Avg Points	Serve Points	Serve Success	Serve Avg Points
715	nan	Marcelo Rodolfo Mendez	COACH	nan	Argentina	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
730	nan	Renan Dal Zotto	COACH	nan	Brazil	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
745	nan	Plamen Konstantinov	COACH	nan	Bulgaria	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
760	nan	Tuomas Sammelvuo	COACH	nan	Canada	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
775	nan	Sheng Wu	COACH	nan	China	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
790	nan	Nicolas Ernesto Vives Coffigny	COACH	nan	Cuba	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
805	nan	Andrea Giani	COACH	nan	France	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
820	nan	Michał Winiarski	COACH	nan	Germany	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
835	nan	Behrouz Ataei Nouri	COACH	nan	Iran	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
850	nan	Ferdinando De Giorgi	COACH	nan	Italy	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
865	nan	Philippe Blain	COACH	nan	Japan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
880	nan	Roberto Piazza	COACH	nan	Netherlands	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
895	nan	Nikola Grbić	COACH	nan	Poland	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
910	nan	Igor Kolaković	COACH	nan	Serbia	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
925	nan	Gheorghe Cretu	COACH	nan	Slovenia	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan
940	nan	John Speraw	COACH	nan	USA	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan	nan

	Country_Name	Mean_Player_Age	Mean_Player_Height	Mean_Player_Total Points	Mean_Player_Average by Match	Mean_Player_Attack Points	Mean_Player_Efficiency	Mean_Player_Attack Avg Points	Mean_Player_Block Points	Mean_Player_Block Success	Mean_Player_Block Avg Points	Mean_Player_Serve Points	Mean_Player_Serve Success	Mean_Player_Serve Avg Points	Std_Player_Age	Std_Player_Height	Std_Player_Total Points	Std_Player_Average by Match	Std_Player_Attack Points	Std_Player_Efficiency	Std_Player_Attack Avg Points	Std_Player_Block Points	Std_Player_Block Success	Std_Player_Block Avg Points	Std_Player_Serve Points	Std_Player_Serve Success	Std_Player_Serve Avg Points
0	ARG	26.857143	193.928571	61.785714	4.792143	48.642857	0.458286	3.759286	8.928571	0.148236	0.699286	4.214286	0.038557	0.333571	3.799942	9.008238	66.079414	5.055095	52.947880	0.246710	4.059089	13.538914	0.100561	1.039256	5.451585	0.043315	0.419571
1	BRA	30.214286	197.714286	56.785714	4.707143	45.928571	0.445479	3.822143	7.142857	0.120350	0.585714	3.714286	0.029679	0.300714	4.806703	8.128095	61.058835	4.566132	53.136203	0.215811	3.967986	6.619983	0.102686	0.521207	5.209881	0.030123	0.405415
2	BUL	29.357143	199.142857	37.714286	3.270714	29.785714	0.338650	2.580714	5.928571	0.147214	0.512857	2.000000	0.024464	0.178571	5.772710	7.471558	39.043228	3.170104	32.503677	0.219640	2.649640	6.366551	0.096838	0.518955	2.385856	0.023804	0.199417
3	CAN	27.000000	197.000000	50.357143	4.286429	40.785714	0.375207	3.461429	5.785714	0.147586	0.490714	3.785714	0.055557	0.332857	3.551814	10.122329	49.156903	4.057767	42.080835	0.217634	3.482806	5.465677	0.131823	0.448853	3.683196	0.066522	0.307882
4	CHN	24.928571	196.928571	47.928571	3.997857	38.214286	0.306050	3.184286	6.142857	0.127479	0.512143	3.571429	0.039043	0.300000	1.979288	7.946594	57.026936	4.750299	47.743189	0.245245	3.978397	7.326004	0.111992	0.610009	4.327271	0.036680	0.359936
5	CUB	25.571429	180.857143	52.714286	4.394286	41.928571	0.325936	3.492857	7.142857	0.201921	0.596429	3.642857	0.025886	0.302857	3.030976	51.942064	68.499779	5.708056	58.827602	0.246434	4.901916	8.008238	0.248412	0.668519	6.511613	0.034204	0.542280
6	FRA	29.500000	196.714286	50.214286	5.340000	41.285714	0.447579	4.375714	5.571429	0.156043	0.630000	3.357143	0.039550	0.335000	2.875627	7.579614	36.008927	3.798099	32.544475	0.216787	3.435549	4.219265	0.098200	0.524126	2.648865	0.029425	0.247814
7	GER	27.285714	198.000000	46.357143	4.250000	37.428571	0.467079	3.476429	5.428571	0.163943	0.463571	3.500000	0.039550	0.310000	4.598136	8.691287	46.521471	3.913965	40.718006	0.222380	3.516057	4.432646	0.090596	0.362419	4.127767	0.035867	0.337434
8	IRI	25.642857	198.357143	33.714286	3.386429	25.857143	0.373057	2.682143	5.928571	0.127957	0.536429	1.928571	0.019743	0.167857	4.955295	9.596531	35.519071	3.499364	27.188394	0.224091	2.911955	10.087801	0.119090	0.839603	2.525692	0.024903	0.218075
9	ITA	24.714286	197.928571	61.857143	5.145000	46.857143	0.394893	3.856429	9.285714	0.183379	0.811429	5.714286	0.049836	0.476429	1.898525	8.721894	66.824520	5.197678	54.576491	0.239488	4.173264	9.084463	0.135678	0.814625	6.603362	0.047654	0.512244
10	JPN	26.285714	190.071429	61.642857	4.220000	52.000000	0.401771	3.552857	5.000000	0.072029	0.338571	4.642857	0.039993	0.328571	3.023716	9.926377	86.538215	5.699071	74.712783	0.232717	4.927673	6.621643	0.073098	0.438035	7.055899	0.061177	0.465103
11	NED	28.071429	199.857143	50.857143	4.431429	39.428571	0.395614	3.455000	6.857143	0.188521	0.586429	4.571429	0.045986	0.388571	4.874806	7.979369	62.338847	5.206086	52.188584	0.310333	4.377790	7.882377	0.142103	0.651373	6.947211	0.046490	0.575057
12	POL	29.214286	197.214286	54.071429	5.665000	42.214286	0.452821	4.495000	7.071429	0.153207	0.708571	4.785714	0.046579	0.462857	3.332326	8.505008	41.235654	4.467172	33.925662	0.283377	3.832225	6.330200	0.097175	0.605384	5.056277	0.041307	0.453388
13	SRB	28.214286	199.071429	51.714286	4.537857	40.642857	0.324836	3.580000	7.285714	0.202821	0.625000	3.785714	0.033386	0.331429	4.353223	7.269158	63.190102	5.352176	53.514351	0.248198	4.547542	9.895054	0.258093	0.822396	5.116768	0.036892	0.430828
14	SLO	29.071429	197.714286	55.428571	4.266429	45.142857	0.371129	3.474286	6.571429	0.096579	0.505714	3.714286	0.023050	0.285714	5.180797	8.480100	71.361177	5.488417	61.020623	0.289932	4.692825	8.317861	0.096958	0.639540	5.703286	0.032523	0.437576
15	USA	31.142857	198.714286	62.571429	4.512143	48.285714	0.440507	3.489286	7.928571	0.168557	0.566429	6.357143	0.061771	0.456429	4.400300	9.522674	55.093926	3.796104	46.881895	0.211454	3.240433	6.330200	0.100300	0.427850	5.624222	0.042766	0.381548

	Team1	Team1_Performance_Ranking	Team2	Team2_Performance_Ranking	Results
0	CHN	125.670000	POL	421.130000	2 - 3
1	POL	421.120000	NED	219.210000	3 - 1
2	POL	419.360000	ARG	311.490000	3 - 1
4	POL	418.900000	CAN	216.250000	3 - 2
5	POL	418.430000	BUL	141.060000	3 - 0

	Country_Name	Mean_Player_Age	Mean_Player_Height	Mean_Player_Total Points	Mean_Player_Average by Match	Mean_Player_Attack Points	Mean_Player_Efficiency	Mean_Player_Attack Avg Points	Mean_Player_Block Points	Mean_Player_Block Success	Mean_Player_Block Avg Points	Mean_Player_Serve Points	Mean_Player_Serve Success	Mean_Player_Serve Avg Points	Std_Player_Age	Std_Player_Height	Std_Player_Total Points	Std_Player_Average by Match	Std_Player_Attack Points	Std_Player_Efficiency	Std_Player_Attack Avg Points	Std_Player_Block Points	Std_Player_Block Success	Std_Player_Block Avg Points	Std_Player_Serve Points	Std_Player_Serve Success	Std_Player_Serve Avg Points
0	ARG	28.357143	194.642857	15.214286	5.070714	12.285714	0.314586	4.095000	2.500000	0.133014	0.833571	0.428571	0.016407	0.142143	3.671243	9.017992	14.900586	4.967682	12.694206	0.254487	4.231913	3.156190	0.123170	1.053000	0.646206	0.028271	0.215127
1	BRA	30.642857	183.642857	15.357143	6.000000	12.928571	0.401600	5.000714	1.500000	0.093550	0.618571	0.928571	0.032043	0.380714	5.400956	52.773255	14.663440	5.476758	13.193613	0.271096	4.813336	1.556624	0.093607	0.758367	1.268814	0.036000	0.564153
2	BUL	23.285714	200.071429	11.642857	3.881429	9.142857	0.340579	3.047857	1.428571	0.061993	0.476429	1.071429	0.030036	0.357143	4.530725	8.606476	14.430775	4.810858	12.196198	0.318115	4.065166	2.651974	0.102184	0.884435	1.730464	0.050784	0.577121
3	CAN	27.571429	197.500000	18.714286	4.678571	14.642857	0.383329	3.660714	2.285714	0.159586	0.571429	1.785714	0.046721	0.446429	2.927700	10.522869	23.173023	5.793256	19.649260	0.312579	4.912315	2.785362	0.256843	0.696341	2.259291	0.056048	0.564823
4	CUB	27.071429	195.642857	22.642857	5.660714	17.142857	0.322793	4.285714	3.714286	0.128314	0.928571	1.785714	0.036121	0.446429	4.859001	9.220438	27.286534	6.821634	21.989008	0.269010	5.497252	4.794686	0.151433	1.198671	2.665407	0.045493	0.666352
5	FRA	28.857143	197.571429	13.428571	3.505714	10.142857	0.353050	2.655000	1.785714	0.139857	0.452143	1.500000	0.043286	0.398571	4.276180	8.262187	13.019006	3.439208	10.861769	0.287620	2.864606	2.082106	0.113727	0.518640	2.066212	0.051429	0.553477
6	GER	26.285714	197.142857	12.357143	4.119286	10.285714	0.390821	3.427857	1.357143	0.080893	0.452857	0.714286	0.023093	0.237857	4.140393	8.483986	12.845874	4.281852	10.992505	0.306632	3.663536	2.373156	0.112254	0.792381	1.138729	0.041025	0.379740
7	IRI	25.214286	198.000000	10.857143	4.035714	8.785714	0.288979	3.297143	1.000000	0.056971	0.368571	1.071429	0.047843	0.369286	4.191173	7.666109	12.346642	5.156532	10.526001	0.250747	4.483536	1.568929	0.094029	0.584661	1.141139	0.055442	0.382833
8	ITA	24.714286	196.000000	13.214286	4.405000	9.571429	0.283629	3.190714	2.500000	0.121157	0.832857	1.142857	0.025200	0.381429	3.561084	10.114727	17.572612	5.857476	13.437687	0.303872	4.479161	3.458546	0.153187	1.152999	2.413333	0.048328	0.805690
9	JPN	26.928571	189.785714	18.071429	4.517857	14.714286	0.390321	3.678571	2.000000	0.110314	0.500000	1.357143	0.034571	0.339286	3.950935	10.452667	17.008886	4.252221	14.777470	0.289432	3.694368	2.287087	0.092375	0.571772	2.307418	0.052683	0.576854
10	NED	28.785714	197.928571	15.500000	5.166429	11.642857	0.368943	3.880714	2.214286	0.160443	0.738571	1.642857	0.046129	0.547857	4.154304	6.580441	21.699876	7.234090	19.121503	0.306620	6.373025	2.359223	0.162233	0.787233	2.273836	0.062049	0.757762
11	POL	27.357143	196.928571	13.500000	4.499286	10.071429	0.416500	3.356429	1.785714	0.199879	0.595000	1.642857	0.072071	0.546429	3.650230	8.175614	10.508239	3.503735	8.879226	0.294286	2.959230	1.528125	0.181423	0.508675	2.023217	0.079943	0.673802
12	SRB	29.071429	198.642857	16.285714	5.427857	12.714286	0.239336	4.237857	1.500000	0.087721	0.499286	2.071429	0.041529	0.690714	6.232227	7.889693	21.456319	7.152308	18.078583	0.257374	6.025411	2.210117	0.142094	0.736650	2.786348	0.049684	0.929238
13	SLO	29.357143	198.071429	11.571429	3.857143	9.142857	0.401364	3.047857	1.571429	0.147679	0.522857	0.857143	0.038379	0.285000	4.325366	7.457204	8.206135	2.736161	7.283883	0.230253	2.428802	2.064882	0.155208	0.687587	1.099450	0.047330	0.366622
14	TUR	28.000000	200.285714	12.785714	4.773571	9.285714	0.480543	3.451429	2.214286	0.167971	0.833571	1.285714	0.040221	0.487143	5.334936	7.700435	14.864405	5.990584	11.275764	0.297310	4.347574	2.636389	0.177059	1.138039	1.728876	0.053064	0.735019
15	USA	26.285714	194.571429	10.857143	3.618571	8.928571	0.419500	2.976429	0.928571	0.068121	0.310000	1.000000	0.040379	0.332857	2.614415	9.621405	12.177262	4.058556	10.314025	0.309003	3.439055	1.268814	0.099823	0.423175	1.617215	0.048695	0.539393

Team 1	Team 2	Prediction Team Won	Prediction Set Score	Actual Team Won	Actual Set Score	Match Outcome
ARG	USA	ARG	3-0	USA	0-3	❌
CAN	CUBA	CAN	3-0	CAN	3-1	✅
SLO	TUR	SLO	3-0	SLO	3-0	✅✅
GER	JPN	GER	3-3	JPN	2-3	❌

First set - winner
Slovenia	1.16	Türkiye	5.0
Second set - winner
Slovenia	1.7	Türkiye	3.0
Third set - winner
Slovenia	1.8	Türkiye	2.85
Fourth set - winner
Slovenia	1.9	Türkiye	2.5
Fifth set - winner
Slovenia	2.0	Türkiye	1.9

Outcome		Prediction Probability
3 - 0		43%
3 - 1		21%
3 - 2		20%
0 - 3		13%
Other		3%

Abstract

Introduction

Volleyball Nations League

Volleyball in the Philippines

Problem Statement and Motivation

Sports Analytics

Industry Analysis – Sports Betting Industry

Integrating Machine Learning in The Sports Betting Industry

Methodology

Scope and Limitation

Scope¶

Limitations¶

List of Tables

List of Figures

Datasets

2023 Volleyball Nations League (Men)

Sources

Volleyball 2023 Standings Dataset

Volleyball Rosters (Men)

2023 Volleyball Match Data (Men)

Exploratory Data Analysis and Preprocessing

2023 Volleyball Nations League (Men)

Table 4.0: Volleyball Nations League 2023 Standings

Table 5.0: Summary Statistics of Volleyball 2023 Standings Dataset

Table 5.1: Volleyball 2023 Standings Dataset - Data Types

Table 5.2: Volleyball 2023 Standings Dataset - Null Values Counts

Table 5.3: Top 5 Teams based on Points

💡 OBSERVATIONS and INSIGHTS¶

2023 Volleyball Rosters (Men)

Table 6.0: Volleyball 2023 Rosters (Men)

Table 6.1: Summary Statistics of Volleyball Rosters 2023 Dataset

Table 6.2: Volleyball Rosters 2023 Dataset - Data Types

⚠ POTENTIAL ISSUES¶

Table 6.3: Volleyball Rosters 2023 Dataset - Fixed Data Types

Table 6.4: Fixed Summary Statistics of Volleyball Rosters 2023 Dataset

Table 6.5: Volleyball Rosters 2023 Dataset - Null Values Counts

Table 6.6: Volleyball Rosters 2023 Dataset - Null Data

TEAM PROFILES¶

Insights on the histograms of player profiles¶

CREATING TEAM PROFILES¶

Table 7.0: Volleyball 2023 Team Country Profiles

Volleyball Match Data (Men)

Table 8.0: Volleyball 2023 Matches

Match Results Distribution:¶

💡 INSIGHTS:¶

Machine Learning

Aggregating Data

Table 9.0: Aggregated Matches

Model Training and Validation

Grid Search (Win Outcome)

Grid Search (Set Outcome)

Interpretability

2024 Predictions

Sample Match

Model Explainability

Figure 22.0: LIME Explanation for Features

RESULTS

Testing June 5 Matches

Figure 25.0: Explanation for Features - ARG vs USA

LIME Explanation for Top 5 Features

SHAP Explanation Top 5 Features

Interpretation:¶

Figure 26.0: Explanation for Features - CAN vs CUB

LIME Explanation for Top 5 Features

SHAP Explanation Top 5 Features

Interpretation:¶

Figure 27.0: Explanation for Features - SLO vs TUR

LIME Explanation for Top 5 Features

SHAP Explanation Top 5 Features

Interpretation:¶

Interpretation:¶

Concluding Interpretation:¶

Linking Bet Smart and Betting Odds

Recommendations

Business Value

Conclusion

References